traceoverlay

TraceOverlay

go get github.com/carbocation/traceoverlay

Purpose

TraceOverlay is a tool for labeling pixels within an image with semantically meaningful labels (semantic segmentation). The tool was written to create training data for aortic segmentation in the following manuscript:

Pirruccello, J.P., Chaffin, M.D., Chou, E.L. et al. Deep learning enables genetic analysis of the human thoracic aorta. Nat Genet 54, 40–51 (2022). https://doi.org/10.1038/s41588-021-00962-4

Installation

cd cmd/traceoverlay
go install
cd ../../

Demos

Demo config files have relative paths that will only work if run from this folder (the same folder as this README).

Running the basic demo

traceoverlay -config demo/basic/config.json

Then navigate in your browser to http://localhost:9019

Running the advanced demo

traceoverlay -config demo/advanced/config.json

Then navigate in your browser to http://localhost:9019

The advanced demo uses a manifest.tsv file to specify which images should be listed (rather than listing all images in a folder).

Demo image licensing

Note: images are used with permission from the "Pexels License" from https://www.pexels.com/photo/christmas-cookies-on-tray-3370704/

Output

Each traced overlay is output as the input filename with .mask.png appended.

Screenshots